Key Takeaways
- Your ledger shows how a customer pays you; a bureau score shows how it pays everyone.
- Eight signals, built from five fields per invoice, cover what a limit review needs.
- Late payments affect seven in ten North American firms and 23% of their B2B receivables (Atradius, 2026).
- 50% of European businesses now ask customers to prepay, up from 46% in 2025 (Intrum, 2026).
- Transformance does not score credit or set limits: Customer 360 holds each customer’s payment behaviour, Vero surfaces changes with evidence, and a person decides.
In This Article
- Key Takeaways
- What does it mean to set credit limits from payment behaviour?
- Which payment behaviour signals should drive a credit limit review?
- How do you calculate each signal from your AR data and payment history?
- How do you turn the signals into a credit limit decision?
- Illustrative example: a growing customer whose limit no longer fits
- How often should you review credit limits, and what should trigger a review?
- What mistakes make credit limits drift away from reality?
- How does Customer 360 support behaviour-based credit reviews?
What does it mean to set credit limits from payment behaviour?
Setting credit limits from payment behaviour means using how a customer actually pays you, such as days beyond terms and broken promises, as the main input for each review.
A behaviour-based credit limit is a limit that is reviewed against the customer’s own payment record with you, instead of being set once from an outside credit risk score. Your ledger also shows whether that record is changing.
US suppliers sell 45% of their B2B sales on credit terms (Atradius, 2026). Across North America, delayed payments affect seven in ten firms and 23% of B2B receivables, and most suppliers put them down to customer liquidity rather than disputes.
In Western Europe, most suppliers “start with internal measures, such as active credit control, payment monitoring, and selective limits by customers” (Atradius, 2026). 39% of European businesses run credit checks, up from 37% in 2025, in a survey of 8,385 businesses in 20 countries (Intrum, 2026).
The sizing formula is on the credit limit page.
Which payment behaviour signals should drive a credit limit review?
Eight ledger signals drive a behaviour-based review: days beyond terms trend, payment-day consistency, on-time share, broken promises, disputes and deductions, partial payments, payer changes and limit utilisation.
Read every signal against the customer’s own baseline, never the portfolio average: a steady payer at 12 days beyond terms is a known quantity, an on-time payer drifting to 9 is news. Thresholds are starting points.
A static limit hides order growth next to slower payment: a winner in the sales report, stretched in the ledger. Never read utilisation alone.
How do you calculate each signal from your AR data and payment history?
Each signal needs five fields per invoice: due date, clearing date, amount, payer and any dispute or promise flag, computed over a 90-day window and a 12-month baseline.
- Dates. In SAP FI-AR the clearing date is AUGDT on the customer line item. It records when the item was cleared in the ledger, so a cash application backlog pushes it past the day the money arrived; where the two differ, use the bank value date.
- Amount. Weight days beyond terms by invoice amount, so one small late invoice does not move the signal.
- Payer. Store the paying account next to the sold-to account.
- Flags. Dispute status, plus the date and amount of every promise.
Freeze the baseline at the last review’s 12 months. A rolling baseline absorbs a slow drift month by month, and the trigger never fires.
Count invoices in an open dispute under disputes, not days beyond terms, so a pricing argument does not look like a liquidity problem. Dispute management sets that flag; see AI dispute management in AR.
The most common failure sits upstream: unapplied partial payments and on-account cash make a good payer look late. Clean up cash application first; see how to automate cash application with AI.
How do you turn the signals into a credit limit decision?
Turn signals into a decision in seven steps: start from your policy limit, compute signals, compare with the customer’s baseline, add group exposure, explain the drift, decide, then record.
- Start from the limit your credit policy gives the customer’s tier.
- Compute the eight signals once cash application is clean.
- Compare each signal with the customer’s own baseline.
- Add up group exposure: open items and open orders across every entity and payer under the same parent.
- Separate explained drift (an open dispute, a seasonal pattern, an announced payer change) from unexplained drift.
- Decide: raise, hold, reduce, or secure with prepayment or shorter terms. Credit decides; sales is consulted.
- Record owner, reason, evidence and the next review date, including every sales override.
Step 4 is where reviews go wrong: with a limit per legal entity, no record shows what the group owes you (see AR across legal entities).
In step 6, a credit hold stops orders, while shorter payment terms or prepayment on large orders keep trade going at lower risk.
Illustrative example: a growing customer whose limit no longer fits
An illustrative distributor on net 30 went from 4 to 17 days beyond terms in under a year as orders rose 30%, so its limit no longer fit.
Companies and figures are fictional. Distributor B buys through two entities under one parent. Its EUR 250,000 limit was set in 2024 and confirmed in April 2026, against a baseline of 4.5 days beyond terms. It then drifted: 8 days in May, 12 in July, 17 in October. Sales now asks for EUR 350,000.
In October the signals read:
- Utilisation: EUR 190,000 open items plus EUR 40,000 open orders, 92% of the limit.
- On-time share: down from 82% to 51% of invoice amount.
- Promises: two of the last three broken, the second in July.
- Consistency: payment-day standard deviation up from 3 days to 11.
- Group exposure: the sister entity has a EUR 150,000 limit with EUR 130,000 open, so the group uses EUR 360,000 of EUR 400,000, invisible on either record alone.
Of EUR 68,000 overdue, an open pricing dispute explains EUR 18,000; the other EUR 50,000 has no reason on file.
The review trigger fired in July, on the second broken promise. The days-beyond-terms rule fired only in October, when August to October averaged 15.3 days against the 4.5-day baseline, three months later.
A credit team might decide: no increase, prepayment on orders over EUR 40,000, a re-review in 30 days, reason and owner logged.
How often should you review credit limits, and what should trigger a review?
Review your largest exposures monthly, standard accounts quarterly and small accounts yearly, and trigger an extra review whenever a signal moves outside the customer’s own baseline.
Set tiers by exposure, not revenue alone. Event triggers to calibrate:
- Days beyond terms 10 or more days above baseline on the 90-day window.
- A second broken promise to pay within six months; see how AI tracks promise to pay.
- A new paying entity, bank account or payment method.
- A sister entity going overdue.
- An AP contact who stops answering. In North America, bad debt is more often linked to customer inactivity or unreachability than to long unpaid large invoices (Atradius, 2026).
EU bankruptcy declarations rose 5.7% in the second quarter of 2026 compared with the first, while business registrations fell 0.5% (Eurostat, 2026). A rise like that pulls reviews forward in exposed sectors; it does not cut every limit.
What mistakes make credit limits drift away from reality?
Credit limits drift when they are set once at onboarding, sized from a bureau score alone, kept per entity instead of per group, or raised without a recorded reason.
- Static limits. The onboarding number survives years of changed behaviour, and good payers get blocked too.
- A bureau score alone. It cannot see the disputes, promises or payer changes on your account.
- Limits per entity. Each record looks fine while the group uses 90% of its combined limits.
- Overrides with no owner. An increase with no reason, owner or expiry date cannot be explained at the next review.
How does Customer 360 support behaviour-based credit reviews?
Customer 360 holds each customer’s payment behaviour in one graph, so Vero from Transformance can surface changes in that behaviour with evidence for the credit team, while a person decides.
Customer 360 is one graph per customer and the working memory of Vero, the platform’s AI agent: payments, promises to pay, disputes and deductions in one record (see the customer 360 view in AR).
When that behaviour changes, Vero surfaces it with the evidence: the invoices, the broken promise, the dispute that explains part of the gap. Vero acts within a 4-level approval model (read, recommend, execute, post) with a full audit trail. Transformance connects to SAP S/4HANA and ECC, Oracle, NetSuite and Dynamics; the limit, the decision and its reason stay with your credit team and in your ERP.
The same history drives collections: see how to prioritise collections with AI.
Conclusion: review limits on behaviour, keep the decision human
Compute eight signals against a frozen baseline, add up the group, separate explained from unexplained drift, and review when the signals move. The decision and its reason stay with a person.
Frequently Asked Questions
How do you set a credit limit for a new customer with no payment history?
Start a new customer on a conservative limit from your credit policy, using financial statements, references and a bureau report, then review it after the first three to six months of payments.
Is a credit bureau score enough to set a credit limit?
A bureau score is a useful outside view, but it shows how a customer pays everyone, while your own ledger shows how it pays you on your terms.
What is days beyond terms, and why does it matter for credit limits?
Days beyond terms is the average number of days a customer pays after the net due date, and its trend against the customer’s own baseline is one of the earliest credit signals in your ledger.
Should a dispute lower a customer’s credit limit?
A valid dispute should not lower a credit limit by itself, because it signals a problem with the invoice, not with the customer’s ability to pay.
Can AI set credit limits automatically?
AI can watch payment behaviour continuously and put the evidence in front of the credit team, but the limit decision should stay with a person. Transformance does not score credit or set limits; Vero surfaces behaviour changes from Customer 360 with evidence, and your credit team decides.
